vaaaaa

1. A Hydroponic Planter System to enable
an Urban Agriculture Service Industry
Akashi Satoh†
†
Department of Communication Engineering and Informatics
The University of Electro-Communications, Chofu, Tokyo, Japan
Abstract— We have developed a compact hydroponic planter
and are conducting cultivation experiments to realize an
agricultural service industry that uses vacant spaces in urban
areas. In order to provide a full-fledged cultivation method for
fruit and vegetables to beginners in agriculture, a low-cost sensor
module and a planter with an integrated remote-control system
have been developed. MQTT, a lightweight protocol for IoT is used
to monitor sensor data and to control a pump. Data is securely
encrypted using TLS/SSL. In addition to monitoring sensor data,
a USB-connected camera-based still-image photography function
with motion detection is included. We are also enhancing the
interfaces to support AI speakers as well as smartphones and PCs.
Keywords—smart agriculture, hydroponic culture, sensor
module, Arduino, Raspberry Pi
I. INTRODUCTION
We have proposed a new consumer-level urban agriculture
methodology by developing and introducing hydroponic
systems for use on rooftops and verandas, as opposed to
traditional agriculture which cultivates the land for production
of crops [1][3]. The systems can be used at office buildings,
hospitals, and elementary schools to provide an oasis, a relaxing
space, and a food education program. We also aim to create new
sixth industries by not only linking producers and consumers,
but by turning the consumers into the producers.
In this recreational urban agriculture, an enhancement of crop
productivity is not important. However, some support systems
for cultivation are required because the main users of our
hydroponic systems are beginners in agriculture. It is not
realistic for a service provider to check each system scattered in
a city directly every day. A remote management system using
IoT technologies, such as an agricultural sensor system for a
large vegetable factory is an effective solution, but it is too
expensive for a small facility in a city area or for personal use.
The sensor system requires high accuracy and high reliability for
optimal control of cultivation environment, which results in a
costly product.
In order to provide a hydroponic system that enables anyone
to casually enjoy cultivation, we have developed monitor and
control system from low-end, high-performance microcomputer
boards (Arduino and Raspberry Pi) and open source software.
Our system is not for mass production, but for recreation where
optimal control with high accuracy sensors is not required.
Therefore our specialized sensor module for hydroponic
cultivation was able to achieve low cost through the use of a
simplified its circuit structure. In this paper, the hardware and
software architectures of our hydroponic culture system are
described.
II. COMPACT HYDROPONIC PLANTER
Fig. 1 shows our compact hydroponic planter and its basic
structure, which is based on a vertical pipe for suspended
cultivation [3]. The simple structure facilitates maintenance
repair and installation, while full-fledged cultivation of fruit and
vegetables can be enjoyed. Water in a tank is pumped up into
two vertical pipes and is showered over the roots of plants put
into the pipes. With no obstacles the rooms extend freely in the
air, and are free from root rot, which is common in hydroponic
cultivation during the summer. Fig. 2 shows tomato cultivation
using the compact planter on the rooftop. Limiting water is
important to increase the sugar content of the tomatoes, and is
easily achieved by controlling the pumping time.
Fig. 1 A compact hydroponic planter for cultivating strawberries
and its basic structure
Fig. 2 Rooftop cultivation of tomatoes.

2. !! # $% 281
Authorized licensed use limited to: Universidad Nacional Pedro Ruiz Gallo. Downloaded on July 03,2022 at 07:57:26 UTC from IEEE Xplore. Restrictions apply.

3. III. SENSOR MODULE
Fig. 3 shows our sensor module for the hydroponic planter.
Liquid fertilizer concentration (based on electrical conductivity
(EC)), water level, water temperature, luminance, temperature,
and humidity are measured by attaching various sensor devices
to the terminal connectors on the module. The water pump and
a fertilizer control equipment are driven by two relay switches.
Fig. 4 shows the compact planter equipped with the sensor
module. The thin, green stick in the tank is a printed wiring
board with electrodes to measure the EC value and water level.
The water circulation pump with hoses and a ball tap for
automatic water supply are placed at the center and on the left,
respectively.
Fig. 3 A sensor module for the hydroponic culture system
stacked on an Arduino-compatible board.
Fig. 4 A compact planter with sensor module.
Fig. 5 is an oscillator circuit for EC measurement whose
frequency varies with solution concentration [2][4]. The
frequency is captured by the frequency counter on the Arduino
ATmega328 processor. In order to prevent buildup on the
electrodes the circuit is active only during measurement. Even
when deactivated, leakage DC current will flow if a voltage
difference occurs between any of the electrodes in the same
water tank. Therefore, a 3-state buffer with a high-impedance
mode was inserted to block the DC flow.
Fig. 5 EC measurement circuit.
Fig. 6 is the water level measurement circuit, where a 3.3-V
source voltage is divided by nine serial resisters. The number of
resisters connected to the GND voltage is changed with the
water level, and the output voltage level is changed between 0
when all resistors are connected to GND and 3.3V when all
resisters are open. The voltage is measured by a 10-bit AD
converter on the ATmega328 processor. This circuit also has 3-
state buffers to prevent DC current flow in the water when
deactivated.
Output
Water
Electrodes
Control
3-state buffer
Fig. 6 Water level measurement circuit.
The water temperature and the luminance are measured
based on the change of resistance of a thermistor and a photo
diode, respectively. A low-cost sensor device [5] is used to
measure atmospheric temperature and humidity. In order to
monitor the status of the water pump and other devices
connected to the sensor module, current monitors are attached
to the two relays, which include a mechanical switch and a
power MOSFET. The sensor module supports Arduino and
other compatible boards as its main board, and we mainly use
an Arduino-compatible board equipped with a micro controller,
ESP-WROOM-32 [6]. The sensor data are sent to the Arduino-
compatible board thorough the serial interface, and are
transferred to a server (Raspberry Pi) by using the ESP-
WROOM-32’s Wi-Fi. In order to reduce sensor cost, the
circuits have simple designs for just measuring frequency and
voltage data, and thus the data needs to be converted to EC,
water level and temperature. As described in the next section,
this conversion is performed by Raspberry Pi, but not by
Arduino.
IV. REMOTE MANAGEMENT SYSTEM
The sensor data are transferred to the Raspberry Pi, which
acts as a sever, over Wi-Fi. The data can be monitored by
portable devices such as smartphones and tablet PCs from a
remote location. One simple solution for this system is that of
installing a Web server on the Raspberry Pi with a global IP
address and transferring the data using the http protocol. If the
Raspberry Pi and the sensor module are not place in the same
LAN, this solution requires to assign another IP address to the
module for the remote pump control.
Therefore, the MQTT (Message Queuing Telemetry
Transport) protocol that does not require a global IP address to
be assigned to the client is used for bidirectional
communication. MQTT, supporting just the minimum
functions for IoT devices, is lighter than HTTP, and thus a low-
price data-only communication service without a global IP,
282
Authorized licensed use limited to: Universidad Nacional Pedro Ruiz Gallo. Downloaded on July 03,2022 at 07:57:26 UTC from IEEE Xplore. Restrictions apply.

4. such as MVNO, can be used. MQTT is more suitable for real-
time operation than HTTP, because MQTT keeps the TCP
session connected once it is established, and thus no hand shake
protocol is required before each data transfer between server
and client. MQTT supports QoS, and three levels 0-2 can be
used depending on importance of data, such as the highest level
3 for the pump control and the lowest level 0 for temperature
that does not change drastically.
MQTT does not support security functionality, such as
encryption. However, cultivation environments of the planter
placed at a private house, such as temperature, humidity, and
illuminance, are private information, and the pump control
should be prevented from being controlled by a malicious third
party. Therefore, TLS/SSL is integrated to encrypt MQTT
packets.
Fig. 7 shows an example network structure of the hydroponic
system. A MQTT client library is installed on the Arduino, and
the sensor data are “published” through a Wi-Fi router. A user
can access the Raspberry Pi which acts as a “Broker” (MQTT
server) using a smartphone and PC to monitor the sensor data,
and to send control data to Arduino which acts as a
“Subscriber” . Node-RED [7] installed on Raspberry Pi is used
as a platform to monitor and control data. Node-RED is a flow-
based programing environment developed by using Node.js,
and provides various applications such as IoT and Web services
by connecting Node modules as shown in Figs. 8 and 9.
MQTT over SSL/TLS
Internet
Raspberry Pi
MQTT Publisher
/Subscriber
Sensor Module + Arduino
MQTT Broker
Fig. 7 An example network structure of the hydroponic system.
Fig. 8 A main flow graph of the hydroponic system.
Fig. 9 Sensor data operation flow of the hydroponic system.
Dashboard [8], one of Node-RED libraries, is used to provide
a GUI, and real-time data monitoring and pump control are
performed on the screen shown in Fig. 10. The graph data on
the Dashboard screen scrolls with time. The data is stored in the
Raspberry Pi server by using the open source database software
MySQL [9] to make them analyzable afterwards. When
abnormal situations, such as the water being empty, pump
failure, and network disturbance over a long period occur a
warning e-mail is sent to a pre-selected address.
Fig. 10 Dashboard for dat monitoring and pump control.
In remote cultivation, monitoring image data as well as
sensor data is important. It is very hard to send hi-resolution
video data in our system, because a low-cost low-bandwidth
network service is used in our system to enable easy enjoyment.
However, plants do not grow rapidly over a short period, and
283
Authorized licensed use limited to: Universidad Nacional Pedro Ruiz Gallo. Downloaded on July 03,2022 at 07:57:26 UTC from IEEE Xplore. Restrictions apply.

5. sending one still image every several minutes to several tens of
minutes is sufficient. On the other hand, it is not favorable if
accidents happening during the interval go unnoticed. In order
to cope with this issue, motion detection by an inter-frame
difference calculation is implemented, and if significant
movement is detected during the regular interval, an additional
image is taken (Fig. 11).
Fig. 11 Camera control menu.
Motion detection and JPEG compression are rather CPU-
heavy operations for Arduino, and thus are performed by the
Raspberry Pi with a general USB camera or an specialized
camera module. A Raspberry Pi 3 equipped with a 64-bit
processor is used for the MQTT server, but performance of the
low-cost Raspberry Pi Zero is also good enough for the image
processing. Currently, captured images are time stamped and
uploaded to a cloud server to be accessed from around the world.
Many pictures are uploaded when the interval time is short
or many motions are detected, and it is difficult to check each
image one by one. Therefore, a method of compressing still
images to a time lapse movie and playing it on a web browser
is being implemented. The movie generation can be executed
on the server, or the Raspberry Pi with the camera can create
and transfer the movie to the server at night when the camera is
not used. In the latter case, it is required to set up a Web server
with a global IP address on the Raspberry Pi to check the video
during the daytime. Therefore, we are also considering using a
free DDNS (Dynamic Domain Name System) service that
connects a host name to a dynamic IP address.
An AI speaker as well as a smartphone and PC are introduced
as an interface of the hydroponic system.
A microphone, a speaker, and LED button of AIY Voice Kit
[10] are added to the server Raspberry Pi to form the AI speaker,
and Japanese speech recognition and voice synthesis are
performed by Julius [11] and Open JTalk [12], respectively.
Currently, basic management functions are implemented such
as reading sensor data and notifying about abnormal states. We
plan to enrich these functions not only for hydroponic
cultivation but also for the user’s day-to-day lives.
V. CONCLUSION
We developed a sensor module for a hydroponic cultivation
system to expand a new type of enjoyable agriculture into urban
areas as a service industry. In order to reduce costs of the system,
general purpose microcomputer boards, Arduino and Raspberry
Pi, are used as an MQTT client and server, and a remote
management system was built using open source libraries and
low-bandwidth communication services. We are also
developing various service businesses for urban smart
agriculture, and expect to report on them in near future.
REFERENCES
[1] Satoh Laboratory, “Urban Smart Agricluture,”
http://satoh.cs.uec.ac.jp/en/research/hydroponics/index.html
[2] T. Nishimura, Y. Okuyama, A. Matsushita, H. Ikeda, and A. Satoh: A
Compact Hardware Design of a Sensor Module forHydroponics, IEEE 5th
Global Conference on Consumer Electronics (GCCE2017), OS-ICE(1)-4
(October 2017).
[3] Ueno Nursery Ltd. , “Easy and Enjoyable Farming”
http://www.uenoengei.com/eng_rakuraku.html
[4] T. Nishimura, et al., “High-Accuracy and Low-Cost Sensor Module for
Hydroponic Culture System,” IEEE GCCE2016, OS-IOT-6, 2016.
[5] Adafruit: DHT11, DHT22 and AM2302 Sensors.
https://learn.adafruit.com/dht/overview
[6] Espressif Systems: ESP-WROOM-32
https://www.espressif.com/en/producttype/esp-wroom-32
[7] Foundation: Node-RED - Flow-based programming for the Internet of
Things.
https://nodered.org/
[8] node-red-dashboard.
https://flows.nodered.org/node/node-red-dashboard
[9] MySQL.
https://www.mysql.com/
[10] AIY Voice Kit.
https://aiyprojects.withgoogle.com/voice
[11] Julius.
http://julius.osdn.jp/
[12] Open JTALK.
http://open-jtalk.sourceforge.net
284
Authorized licensed use limited to: Universidad Nacional Pedro Ruiz Gallo. Downloaded on July 03,2022 at 07:57:26 UTC from IEEE Xplore. Restrictions apply.